Is There Value in Identifying Individual Genetic Predispositions to violence?

Genetic Predispositions to violence?

David Wasserman

n this article I want to ask what we should do, either collectively or individually, if we could identify by I genetic and family profding the 12% of the male popu-

lation likely to commit almost half the violent crime in our society. What if we could identtfy some individuals in that 12% not only at birth, but in utero, or before implantation? I will explain the source of these figures later; for now, I will use them only to provide a concrete example of the kind of predictive claims we can expect to be made with some frequency, and some scientific credibility, over the next generation. I will adopt an outlook that one commen- tator has called “pragmatic optimism,”’ but which could also be called technological optimism - the belief that a science or technology will achieve many or most of its advertised goals. My optimism will be directed towards human behavioral genetics, the source of predictions like the one I just offered; I will assume that this controversial discipline will achieve a substantial pan of its scientific ambition to identlfy genetic differences among individuals that help predict and possibly explain future behavior, psychological health, and cognitive skill.2 This optimism is very limited - it concerns the scientific success of behavioral genetics, not the social value of that success. There is good reason for such limited optimism: if I assumed instead that the human behavioral genetics of the early 21st century would have no greater scientific yield than the eugenic research of the early 20th century, then the only issue to discuss would be the moral and social signifi- cance of the false expectations that discipline would continue to engender.

On the other hand, it would be foolish, and in retro- spect, quite likely embarrassing, to abandon any skepticism

Jot~mzaloflau; Medicine & Etbh, 32 (2004): 24-33. 0 2004 by the American Society of Law, Medicine & Ethics.

about claims of future progress in behavioral genetics, especially if we have reason to suspect that those claims are not always believed even by the researchers who make them.3 Although I am not a scientist, and certainly not a behavioral geneticist, I will suggest that recent research gives us ground for cautious pragmatic optimism. I will also argue, however, that there is little reason for a broader optimism: even if behavioral geneticists do identdy practi- cally as well as statistically significant associations between specific genetic variations and important behavioral, psychological, and cognitive characteristics, associations that hold a m the social environments we are likely to find our- selves in, there may be little we can, or should, do about it.

I will, however, suggest why the findings of the next generation of behavioral genetics may have stronger implications for moral evaluation than for practical intervention. If those findings have roughly the same character as the findings of the Caspi, et. al.4 research, from which I appropriated the percentages at the start of this paper, genetic research will give us ways to think about moral and legal responsibility for violence that are a good deal more nuanced than one might have expected from the past decade’s debate about the social implications of behavioral genetics5

Sumrmc BACKGROUND: A QUICK INTRODUCTION T~HUMANBEHAWO~GENEI~CS First, though, I need to provide a bit of background on the enterprise of behavioral genetics; on the methods by which it seeks evidence of genetic differences that influence complex behaviors and psychological traits, and on the interpretive challenges that evidence confronts. The defining feature of behavioral genetics is its focus on - critics would say its obsession with - the effects of individual

24

7be Journal of Law, Medicine G Ethics

differences in genetic constitution.6 Humans beings share over 98% of their genome with chimpanzees and more with each other; behavioral genetics is concerned with the relatively small portion that differs from person to person. Many critics believe this emphasis is misplaced; they doubt that the effort to explain differences in human behavior and personality even partly in terms of genetic differences is scientifically promising.’ Some critics argue from an evolutionary perspective that while all human beings, or all human males, may well have genetic propensities for violent behavior, individual differences in the strength of those propensities are unlikely to have a genetic s0urce.H And even if genetic differences among individuals made them more or less prone to violent behavior, they are likely to contribute far less to the understanding of human violence than our shared genetic legacy and our environmental differences. Such criticism may increase our skepticism and lower our expectations about finding sigtuficant individual differences in genes associated with behavior. But it cannot preempt the need to assess the specific methods and reported fmdings of recent human behavioral genetics.

Heritability Studies

Until techniques were developed in the 1970s for isolating and manipulating genetic and other molecular material, human behavioral genetics was largely limited to heritability studies. Heritability research takes off from the scarcely debatable observation that many behaviors and psychological traits “travel in families.” But parents usually confound the assessment of genetic influence by providing their children with rearing environments as well as genes; the transmission of behavior from one generation to the next can be attributed to either one.

Human behavioral genetics exploits processes which tend to tease apart these genetic and environmental contri- butions: “identical,” or monozygotic (MZ) twinning, “fraternal,” or dizygotic (DZ) twinning, and ad~pt ion .~ Comparisons of MZ and DZ twins, and of biological and adoptive children and parents, allow researchers to estimate, using a number of debatable assumptions, the proportions of the population variance in the traits studied that can be attributed to genetic and to environmental factors. Thus, a typical finding of a twin study might be that “60% of the variance in IQ score is attributable to heredty.” Fairly high heritabilities have been found for a number of significant behavioral and psychological traits - though interestingly, not for violence.1o

There has been a great deal of debate, much of it among behavioral geneticists, about the validity of the assumptions on which these comparisons rest: e.g., are the rearing environments of DZ twins as much alike as those of MZ twins, or are identical twins environmentally as well as genetically more alike, down to their identical

wardrobes? Does the time adoptees spend with their biological parents help explain the behavioral similarities of the adoptees and their biological parents? What about the adoption agencies’ non-random placement practices?”

But the more important challenge for heritability studies is what one author has referred to as their “explanatory limit^."'^ These studies, however well designed, cannot reveal what is actually inherited, or how whatever is inherited contributes to the development of the traits studied.13 They look only at the population variance of behavioral and psychological characteristics. They attempt to measure the proportion of that variance which cannot be attributed to environmental differences or chance, and must therefore be attributable to genetic differences. But those genetic differences need not be the same for every subgroup, let alone every individual, in the population. The claim, for example, that heredity accounts for 60% of the variance in IQ does not mean that 60% of an individual’s IQ score, or deviation from the mean IQ, is genetically determined (if such claims are even intelligible).14

Moreover, heritability studies cannot investigate or distinguish among possible causal pathways from gene to trait or behavior. Thus, to take an example from Christo- pher Jencks, people with red hair might have below-average IQ solely because they are neglected and mistreated as children throughout the society. l5 Common sense would treat this as an environmental explanation of their lower IQ, but because there would be little or no variation in the treatment of red-haired children across the society, it would be counted as genetic in a heritability study, since there would be little variation in outcome across the environments actually studied. The fact that genes for hair color lower IQ only by evoking discrimination would not be reflected in the proportion of the variance assigned to heredity. The example becomes more familiar, and more troubling, if we substitute dark skin for red hair.

The indirect manner in which behaviors and mental traits and behaviors are typically measured give such alternative explanations further plausibility. If the researcher adopts arrests or convictions as a proxy for criminal behavior or disposition, or scores on a written test as a measure of task proficiency, the causal pathway from the genes to the measured variable may not run through the trait or behavior at all: What may be transmitted genetically is not a propensity to commit crimes, .but a lack of talent for concealment or evasion; not proficiency in a specific cognitive task but a general facility with written tests.

Molecular Behavioral Genetics

For the past twenty years, the technology has been available to overcome one of these explanatory limits; to iden* specific genetic variations that are associated with behavioral and mental traits.16 That technology, however, cannot

25

Volume 32: 1, Spring 2004

by itself overcome the other explanatory limit - it cannot explain the causal role such variations play in producing the behavior in question. As one persistent critic of human behavioral genetics has pointed out, its molecular turn has still yielded only correlations, not casual mechanisms or pathways.” But this may be changing.

In the late 1980s and early 1990s, researchers identi- fied a number of genetic markers, and in some cases, specific genetic variants, or polymorphisms, that were associated with diseases known or suspected by their inheritance patterns to have a substantial genetic component. Behav- ioral geneticists were quick to adopt the same techniques, hoping to replicate the dramatic successes of medical genetics.18 With these techniques, they attempted to associate behavioral and mental traits with specific chromosomal regions or genetic variations, and were soon reporting markers for a number of psychiatric and behavioral conditions, including bipolar disorder, schizophrenia, and alcoholism. Early findings in the f a t two of these areas had to be retracted, however, and findings in the third remain mired in controver~y.’~ In the decade after the first application of molecular genetic techniques to psychiatric and behavioral disorders, not a single consistently replicated, generally accepted association was found between a specific gene or marker and a common behavioral or psychiatric disorder.

In 1993, researchers did find a marker, then a gene, associated with violence and aggression, in the male members of a Dutch family.M This was an unexpected and somewhat awkward finding; it looked like the kind of “major” crime gene that researchers had not been expect- ing to discover. While the family studied was atypical in several relevant respects, the study had enormous impact, in part because the affected gene was know to produce a protein, monoamhe oxidase A (MAOA). MAOA is involved in the metabolism of serotonin, a neurotransmitter thought to play a major role in neuro-chemical regulation, and in mediating between genes and behavior. Serotonin has been implicated in psychiatric and behavioral conditions rang- ing from depression to impulsive violence. Critics argued that the llnk between MAOA and violence and aggression, was, for this very reason, more tenuous than the researchers suggested. Thus, Evan Balaban contended that “the biochemical equivalent of hitting a subject on the head with a club may explain a pattern of pathology in a small number of individuals but will not be very enlightening for either the scientific study of behavioral biology or for general societal problems involving crime and violence.”21 Although some of the men in the family studied engaged in clearly violent and antisocial acts, that conduct may well have reflected a more general deficit. The mutant-MAOA males may not have been more aggressive, but less intelli- gent, lacking constructive outlets for their aggression or clever ways of concealing it. The Brunner findings were

awkward for another reason as well. High MAOA activity had been associated with low serotonin levels, and low serotonin levels with impulsivity and anti-social behavior, among a host of psychiatric and behavioral disorders. Brunner‘s subjects, lacking virtually any MAOA activity, might have been expected to be unusually pacific, not violent.u This does not suggest that Brunner‘s findings were mistaken, but rather that the connections among genes, enzymes, neu- rotrans~tters, and behavior are more complicated, and less easily understd, than researchers might have expected.

The controversy over the interpretation of the Brunner, et. al., findings makes it clear that ideneing specific genes associated with behavioral or psychological traits is only one step toward explaining the ways that genes contribute to those traits. Not only are behavioral and psychological traits likely to be affected by a large number of genes (polygeny), but each of those genes is likely to have effects on many different aspects of development, and affect those traits in a variety of direct or indirect ways (pleiotmpy). Because researchers will rarely be able to track the full range of functions and interactions a single gene or gene product can have, they will have to acquire a great deal of knowledge about developmental and physio- logical processes to MITOW the range of plausible causal pathways. That knowledge can hardly come h m molecular genetics alone.

The Caspi, et. aL S t u d y In 2002, behavioral geneticists were finally able to iden@ fairly common genetic variants that appeared to have an intriguing association with violent behavior.23 Their findings, interestingly, were made in the context of research designed to yield information about a wide array of non-genetic factors shaping behavior and psychology - a longitudinal study of a large cohort of New Zealand youth, begun well before it was possible to identify genetic differences among its subjects. What they found - to oversimplify - was that the only variable studied that had a main effect on violence (which they measured in several ways and by a statistical composite of those ways) was childhood maltreatment, found or suspected for 36% of the cohort. This included parental abuse or neglect, and other forms of harmful treatment by their parents and families.”

The researchers genotyped almost every male mem- ber of the “Dunedin cohort,” looking for common variants of the promoter sequence for the MAOA-linked gene, in contrast to the very rare “null allele” found by Brunner et. al.25 Caspi, et. al., found that roughly a third of the cohort had variants producing low MAOA activity (low MAOA); the rest, variants producing high MAOA activity (high MAOA).26 Low MAOA did not have a significant main effect on violence. But it had a striking interaction with maltreatment - while there was no significant difference

26

Thejournal of Law, Medicine C Ethics

between high and low MAOA subjects who had not been maltreated, there was a substantial difference between those who had. Maltreated subjects with high MAOA levels were no more violent than the average subject, while the 12% of subjects who were both maltreated and low MAOA accounted for fully 44% of all convictions for violent crimes.” Almost a third of the severely maltreated low-MAOA males had convictions for violent crimes by their late 20s. The maltreated low-MAOA males also had significantly higher scores than other subjects on the other measures of violence.28

The researchers suggested that the role of MAOA activity in violence was an indirect one, with high levels protecting against the psychological effects of childhood maltreatment, and low levels making children vulnerable to those effects:

Numerous biological and psychological processes have been put forward to explain why and how experiences of maltreatment are converted into antisocial behavior towards others . . . but there is no conclusive evidence that any of these processes can account for the progression from childhood maltreatment to later criminal violence. . . . The search has focused on social experiences that may protect some children, over- looking a potential protective role of genes.B

The researchers attempted to rule out other possible explanations for the Maltreatment x MAOA interaction: low MAOA activity, as opposed to no MAOA activity, did not appear to produce a general cognitive deficit; low MAOA males did not have lower IQs than high MAOA males. Nor did maltreatment appear to lower MAOA activity: average MAOA activity was no lower for males who were maltreated than for males who were not. Nor did the Maltreatment x MAOA interaction depend on social class: high MAOA activity appeared to confer its protective effect on rich and poor alike.%

There are, undoubtedly, confounding variables and alternative explanations that Caspi, et. al., did not rule out. Moreover, the effects found by Caspi, et. al, like the more pronounced ones found by Brunner, et. al., reversed the association between high MAOA/low serotonin and antisocial behavior (ASB) repeatedly found in other research. Finally, ruling out alternative explanations, however exhaustively, cannot substitute for establishing causal mechanisms for the protective effect of MAOA activity suggested by the researchers. But here I will indulge my technological optimism, and assume that future research fills in the picture of genetic vulnerability that Caspi and his colleagues have sketched. That assumption allows me to ask how their account of the indirect role of low MAOA activity in violence, and the identification of high-risk individuals their research facilitates, should affect social

intervention and moral appraisal. I think it has more promise, or at least fewer pitfalls, for appraisal than for intervention.

OR bKll”G THE EFFECT3 OF INDIVIDUAL GENETIC PREDBIXlSmONS To VIOIRVCE

The value of MAOA genotyping as a basis for violence- reducing interventions will depend on two factors: 1) the role of such genotyping in developing appropriate and effective treatments for individuals; 2) the importance of individual differences in violence-propensity for the incidence of, and the harm caused by, violent behavior.

Identification and Treatment The recent history of medical genetics should make us pessimistic. The identification of predisposed individuals, and even of specific alleles, has done little to facilitate treatment. The discoveries of mutations associated with Cystic Fibrosis and Huntington disease are prime examples of successful genetic identifications that have yielded less thera- peutic benefit than expected. But there may be a wider range of interventions available for behavioral than disease predispositions. Contrary to a fallacious but widespread assumption, the “treatment” of a genetic predisposition to violence can be environmental rather than medical or pharmacological. In light of the respective roles of environments and genes suggested by the Caspi, et. al. study,31 environmental interventions might seem more likely to be effective and acceptable than medical ones. But as I will argue, that appearance may be deceptive.

Initially, it might seem more acceptable morally to break a cycle of violence in a family of low-MAOA members by strategies to reduce parents’ maltreatment, such as anger management training, than by strategies to increase children’s resistance to its effects, such as medical interventions to increase protective MAOA activity, even if the latter were more effective than the former. We can find an analogue in the choice between interventions to reduce the harmful effects of other toxic environments, for example, by reducing pollution or by increasing immunity to its effects. The moral case for one kind of intervention over the other is even stronger here, because maltreatment would be objectionable even if it didn’t increase the predisposition to violence, while pollution is objectionable mainly because of its health and environmental effects; indeed, those effects may be part of what it means to call a discharge or emission “polluting.” Despite this initial preference for environmental interventions, however, the difficulties raised by their real-world implementation should reduce their appeal.

Given the dismal record of child protection agencies in preventing even the worst forms of abuse and neglect,32

27


it is doubtful that preventative intervention involving state supervision or removal of maltreated children would be very effective. Moreover, such intervention would only be feasible if there were a lower threshold for state intervention in families than the prevailing abuse or neglect standard, since even “severe maltreatment” as defined by Caspi, et. al., does not require legal abuse or negle~t?~ The threshold for intervention would either have to be lowered for all families, or lowered selectively for families with low-MAOA children. That latter, however, would require the prior identification of low-MAOA children, as well as the judicial recognition of low MAOA as a risk factor warranting heightened surveillance and lower thresholds for intervention in potentially abusive families.

Both these alternatives - lowering the bar for intervention generally, or just for families with low MAOA children, would be politically, legally, and morally problematic. The abuse and neglect standard represents a delicate compromise between parental autonomy and child welfare, which it would be risky to abandon.% Reducing the threshold for intervention would pose a further threat to the autonomy of the poor and working class parents who now fall under state ~urveillance,’~ while targeting low-MAOA children would significantly infringe the privacy of all children by requiring their genetic screening for non-medical purposes. Moreover, a selective lowering of the threshold for intervention would present the unseemly spectacle of a government displaying greater solicitude for potentially violent children, allowing equally maltreated high-MAOA children to languish in dysfunc- tional families because their malmatment posed little danger to society. Of course, some would argue that the operation of the abuse and neglect system already presents the unseemly spectacle of state agencies removing children because of parenting difficulties often directly attributable to state neglect of poor families.% But it would only heighten the injustice to impose heightened scrutiny on poor families with children who were genetically vulnerable to the adverse behavioral effects of maltreatment.

The heightened surveillance of, and intervention in, families with low MAOA male children would be especially troubling if they were directed disproportionately at lower income or minority families. Reassuringly, Caspi, et. al. found the same Maltreatment x MAOA interaction when they controlled for social class.” They did not, however, report on the frequency of “probable” or severe maltreatment% by social class. It would not be surprising if that proportion were higher for lower class and minority subjects, in part because poor and minority families are more exposed to state surveillance and intervention regardless of their propensity to mistreat their children or the genetic vulnerability of those children, in part because severe poverty makes neglect harder to avoid?9 Although making a charge of abuse or neglect a prerequisite for

mandatory genetic testing of children for MAOA activity would alleviate some civil libertarian concerns, it might only exacerbate class bias.

And these are merely some of the more benign possibilities. A government willing to subject a much of its population to mandatory genetic screening for non-medical purposes might well engage in harsher preventative practices. It might subject those with “individual vulner- abilities” to violence - Fred Goodwin’s infamous phrasea - to heightened surveillance, forced medication, or even preventative detention. I will not dwell on these possibilities, not because I think they are fanciful, but because I do not think I need to make the case against them. My point has been that even more restrained and humane interventions may be deeply problematic.

Such considerations reduce the comparative moral and practical appeal of environmental over biomedical interventions. Caspi and his colleagues do not even mention environmental interventions, they assert only that their work “could inform the development of future pharmacological treatment^."^^ Pharmacological treatments, however, such as the injection of MAOA protein or the viral transfer of high- MAOA pmmoter sequences, might not be effective, and they would be problematic even if they were: they would still require widespread or mandatory screening, and they would “immunize” bad parents against the social consequences of their harmful conduct as well as protecting their children from the harmful effects of that conduct. It would be tempt- ing for the state to set a low threshold for pharmacological prevention for large categories of potentially vulnerable young men. For example “prescribing” MAOA supplements for all young men screened as low MAOA, regardless of social class or family circumstances, would avoid blatant class bias, as well as intrusive and expensive inquires into maltreatment than fell short of abuse and neglect. Yet such dragnet screening and wholesale prescription would raise a host of safety and civil liberties concerns.

More troubling still would be policies promoting the selective implantation of embryos carrying high-MAOA alleles in women susceptible to abusive parenting, or the selective abortion of fetuses carrying low-MAOA alleles. Preventing the birth of low MAOA children through pre-implantation genetic diagnosis (PGD) or pre-natal testing ( P W would provoke resistance even among those generally accepting of those forms of testing and selection, because parents would appear to be selecting children they could mistreat with relative impunity. Would germ- line genetic engineering to ensure high MAOA production be more acceptable? Such engineering could, to the extent it was identity-preserving, be viewed as a preventative measure like immunization, in the face of a “pathogenic” environment. But viewing such genetic engineering as a preventative measure would seem disingenuous when the future child’s parents were both the engineers and the

28

‘I;beJournal of Law, Medicine & Ethics

“pathogens.” Such genetic engineering by a potentially abusive parent would bear a disturbing resemblance to genetic screening by an empIoyer seeking to hire employ- ees immune to the adverse health effects of its toxic workplace. It would also create a sigmficant “moral haz- ard,” if parents who immunized their children felt less pressure to restrain their own conduct. Even the most enthusiastic proponents of pharmacological prevention would not claim that it could fully protect children from, or fully compensate them for, bad, deficient, neglectful, or abusive parenting.

Reducing Violence by Identifying

As to the second factor I mentioned above - the impact on levels of violence of identlfying and treating males predisposed to violence by maltreatment and genetically low MAOA - a lot would depend on social context. In a generally pacific society, with low rates of homicide, robbery, rape, and other assaults, identifying and treating maltreated males with genetically low MAOA activity - if it could be done, and done morally - might signdkantly reduce the already low levels of crime (although the detection of its impact would obviously be blunted by the

Genetically-Lo~ MAOA Males

for example, expect that low MAOA men, perhaps even those who were not subject to maltreatment, would be more likely to engage in appropriate defensive violence, combat heroics, and life-saving bystander intervention. Maybe Kitty Genovese, who was murdered on a New York City street while more than thuty witnesses did nothing, would have been more likely to survive her attack if there were more low MAOA men on her block. (Such benign effects might be harder to assess, however since they may be less frequent and less likely to leave an official record.) If low-MAOA activity increases the odds of appropriate as well as inappropriate violence, its genetic or pharmacc- logical control may provoke a debate similar to one of those about gun Do we wish to pharmacologi- cally “disarm” a large proportion of our citizenry to reduce violence, when doing so will also reduce our resources for combating personal violence, foreign aggression, and physi- cal disaster? Or is it naive to think that the violent and antisocial predispositions of Caspi’s maltreated low MAOA subjects would have the same or similar neurophysiology as predispositions to self-defense, the defense of others, or the rescue of bystanders and victims? It may no less facile to assume that every criminal is a potential cop than to assume that every cop is a potential criminal.

low “floor”). Even in such generally pacific societies, there will always be maltreated children, and reducing maltreat- THE M~~ AND AP~RA~SAL bmum ment might have the happy effect of reducing the violence ~m~~ ~ R E D I ~ ~ ~ ~ E D T~ vIoma it generated.

But such efforts might seem trivial or irrelevant in per- vasively violent societies. In Bosnia, Kosovo, Rwanda, Northern Ireland, Kashmir, Afghanistan, Israel and Pales- tine, or Sierra Leone and Liberia, it would appear futile to identdy the third of the male population that had low MAOA activity; as futile as it would have been in the First or Second World War, where a majority of the young male population of the belligerent countries was engaged in some form of violence. In societies where violence is orga- nized, systematic, and pervasive, MAOA activity may explain very little about the behavior of the foot soldiers. Even in settings where the level of violence is less extreme, ideo- logical, political, and religious violence may be better explained by propensities to obey authority and “authoritarian personalities,”42 which may or may not be usefully attributed to genetic variations. (Perhaps in such settings, Caspi’s findings4) would not be replicated, since maltreatment and MAOA activity would both play a far less important role in the production of violence.) Any genetic factors that differentially predispose men toward violence are likely to matter in far less settings where violence is pervasive than in settings where it is rare or occasional.

A further problem with genetic or pharmacological prevention is that there may be other effects of low MAOA activity we would be more hesitant to suppress. We might,

Thus far, I have argued that there would be abundant reason for caution about doing anythtng with the finding that a si@icant proportion of men may be vulnerable to the violence-promoting effects of childhood maltreatment. I’ve suggested substantial practical and moral barriers to the prospective use of such genetic information. I now want to suggest that there may be more value in the retrospective use of such information - in the moral, and possibly legal, assessment of, and response to, genetically low MAOA males who commit violence after a childhood of maltreatment.

One hardly needs to be a genetic determinist, who holds that genes fully determine behavioral and other to believe that credible evidence of genetic influence on behavior should affect the moral and possibly legal appraisal of that behavior. For those who believe that a complete causal explanation of human behavior is incompatible with the attribution of blame and the imposition of punishment, credible evidence of the role played by genetic variations would fill in impomnt causal gaps in that explanation, making the prospect of a complete explanation appear more likely, and its completion appear closer. Such evidence would thereby threaten to undermine the practices of blaming and But few laypeople, and only a minority of moral philosophers, may be convinced that determinism is incompatible with the attribution of responsibility; most laypeople and

29


philosophers may be what philosophers call “compatibilists,” inclined to believe that causal determination alone is not incompatible with moral responsibility, and that difFerent causal pathways may have different relevance for the assignment of responsibility.

To oversimplify a discussion that has grown formida- bly complex in recent years:’ many compatibilists find one, or both, of two considerations relevant to the assignment of moral responsibility, regardless of whether the agent’s acts are fully determined.

The first is the extent that the agent acts as the result of normal deliberative processes. Thus, an agent who robs banks because he believes that that’s where the money is and wants the money is more responsible for his actions than an agent who robs banks because he cannot understand the concept of private ownership, has a delusional belief that the money in the bank already belongs to him, or literally cannot resist the impulse to rob banks, though he does not want to and considers it both wrong and unwise. Some compatibilists hold that agents are no less responsible by virtue of having strong desires and weak self-restraint, unless they experience a breakdown in their practical rationality;@ others argue that agents who have acquired, through no fault of their own, strong impulses or weak self-restraint, are less blameworthy for their bad actions, even if they are still partially re~ponsible.~~

The second feature is the extent to which the agent identifies with, or endorses, the desires on which he acts. On the analysis of moral responsibility introduced by Harry Frankfurt a generation ago, agents are morally responsible to the extent they identify with, accept, or ratify, the desires on which they An agent who acts on desires she on some level rejects (although she acts freely) does not have a free will, and is not fully responsible, morally, for her actions. Even if the agent’s identification with the desire on which she acts is as fully determined as the desire itself, those acting on desires they embrace act more “wholeheartedly,” and seem more responsible for their actions, than those who are ambivalent or unhappy about the desires on which they act. These more nuanced accounts of the attribution of responsibility (each of which have many variants I cannot attempt to even sketch),5I may give claims of complex genetic influence like Caspi)s,5* and the causal accounts they are likely to generate, much more of a role than incompatiblist accounts that hold people responsible if and only if their behavior is not causally determined.

Clearly, evidence of a genetic vulnerability to the violence-provoking effects of childhood maltreatment would have varying relevance on different compatibilist accounts. It would be less likely to mitigate, let alone exculpate, on accounts which regarded agents as responsible unless their normal deliberative processes were disrupted or bypassed.53 It would more likely be relevant

on more lenient accounts, that regarded factors making it difficult but not impossible to exercise self-restraint as mitigating, if not exculpating, the agent’s conduct.54

As I’ve noted, however, Caspi and his colleagues can only speculate about the mechanism by which low MAOA activity facilitates or permits childhood maltreatment to promote violence in its victims. A close analysis of how the genetic vulnerability they find may be relevant on various compatibilist accounts will have to await the discovery of such mechanisms. But even without knowledge about specific mechanisms, claims of MAOA may have some relevance for the moral and legal appraisal of those who commit acts of violence.

The most obvious way in which evidence of a “Caspi- predisposition” to violence could affect the moral or legal appraisal of the agent is in offering a quick, if incomplete, rebuttal to the standard response of sentencing judges to pleas for mitigation based upon childhood maltreatment: that the overwhelming majority of maltreated children, like children from rotten social backgrounds, end up as law abiding citizens.55 What should we expect any less of this maltreated defendant? Such a question invites evidence of differentiating factors such as genetically low MAOA. In peaceful Dunedin, a large minority of maltreated males with low MAOA did commit (or at least were convicted of committing) violent offenses;” in the violence-breeding hothouses of American cities, it might well be a majority, even a large majority. We can also imagine behavioral genetic research finding additional alleles, besides those regulating MAOA activity, operating synergistically with maltreatment to promote violence. A defendant able to adduce such additional genetic factors might well find him- self in a (statistical) cell in which 80% or 90% of the individuals committed violent acts. Although the moral relevance of the claim that “almost everyone in my cell does it” is not obvious, especially in the absence of a causal explanation of why they do it, or are predisposed to do it?’ such figures would seem to shift the burden of explanation to the 10% or 20% who didn’t act violently (or weren’t caught at it).

Moreover, the research I’ve been discussing does suggest a causal mechanism, however sketchily: that high levels of MA0 protect or insulate an individual from the violent- and anti-social effects of sustained maltreatment; that they give the individual a resilience, or equanimity, in the face of environmental stress, that low-MAOA males lack. The neurochemical or cognitive mechanisms by this deficit would promote violence have yet to be suggested, let along explained, but I have been taking the technologi- cally optimistic view that a scientifically plausible explanation will be forthcoming.

I believe that the interactive character of the effect of MAOA activity on violence, together with the kind of causal mechanism that’s been suggested, would support stronger

30

The Journal of Law, Medicine & Ethics

mitigation on some compatibilist than would evidence of a “simple” genetic predisposition to violence, e.g., a genetic variant with a violence-predisposing effect in all known environments. The latter predisposition, however striking, would only add to the roster of predisposing factors that provoke the ambivalent response I outlined in an earlier paper? It would soften one type or aspect of moral disap- proval, accountability - the extent to which the individual is held responsible for violating norms of interpersonal behavior. But it would sharpen another aspect - attributability, the extent to which those violations, however voluntary, are regarded as truly his, as “in character.” Evidence of violent character or predisposition would thus have opposing effects on what Gary Watson calls “the two faces of re~ponsibility.”~’ Those with a simple genetic predisposition to violence would find some mitigation for violent acts in their possession of an allele that made it more difficult to exercise self-control, but possessing such an allele might also make them appear to be inherently violent or impulsive. The only way in which the identifka- tion of such a genetic “main effect” would be likely to have an exculpatory or mitigating effect on balance would be if it revealed a causal mechanism that disrupted normal processes of practical decision making or voluntary action. Such wholesale disruption would satisfy the stringent requirements for exculpation of compatibilists like Morse.m But if a genetic variant produced such an extreme disruption of practical reason or self-control, evidence of the genetic origins of that disruption would hardly be needed for mitigation.

In contrast, the discovery of a genetic vulnerability to environmental stress might well decrease accountability without increasing attributability. A person lacking genetic protection from violence-provoking abuse might not seem inherently violent, in contrast to a person genetically predisposed to act violently in response to ordinary stress and provocation. It may be easier to see the former than the latter as the victim of bad “moral luck6’ - not the “constitutive” moral luck of a violent genetic predisposition virtually certain to be triggered by ordinary normal stress and provocation, but the conventional bad luck of encountering extraordinary stress or provocation. The low MAOA individual would appear to be doubly victimized - by maltreatment from his parents or caregivers, and by a lack of genetic protection from that maltreatment. More- over, it would be difficult to even characterize such an individual as “genetically predisposed to violence,” since he would possess a genetic variant that contributed to violence only in depriving him of internal protection from the damaging impact of childhood maltreatment.

Of course, the extent to which the mitigating impact of such complex genetic predispositions exceeds their aggravating impact will depend very much on the details of the causal mechanisms through which low h4AOA activity

mediates the impact of childhood maltreatment on violence. Those details, of course, have yet to be supplied, and Caspi’s findings seem inconsistent with what is known about the relationship between MAOA, serotonin, and violence.62 But that should not keep us from speculating. To Patricia Greenspan, writing several years before the Caspi research, a breakdown or deficiency in the inhibitory system (in which serotonin is involved) suggested a different kind of excuse than the more familiar “irresistible impulse”:

The point is not that genetic explanation of criminal behavior is itself problematic for free will but rather that it may pose problems in combination with a shift way from our intuitive view of the causal mechanism involved in failures of self-control. One might say that the current research model takes the “mania” out of kleptomania - the sug- gestion of “irresistible impulse” as a stirred up counterforce to self-control, like a demon fight- ing oneself- and substitutes personal inadequacy, a shortage of the means of self-control.. , . The cause on this account, rather than shortcutting the will, shortchanges it: it denies the agent the very stuff of self-control, on the model of a spy under torture who just becomes too exhausted to resist after a long period of sleep deprivation.. . . The impulsive criminal has the same impulses as the rest of us, at least initially, on this account; he simply lacks our apacity to suppress them, or to let them pass ~nsatisfied.~3

Greenspan acknowledges that “such a simple account of the causal mechanism is [not] adequate to capture it.”@ That account also has uncertain implications for moral appraisal. On theories of moral responsibility that make the agent’s identification with his desires critical to his responsibility for acting on “inherited impulsivity” will be less mitigating than “irresistible impulses” to the extent that the agent is more likely to identify with familiar impulses than the powerful, opaque ones that characterize many compulsions and manias. And on theories that see moral responsibility as reduced more by mechanisms that “shortcut” than “shortchange” the will,% impulses powerful enough to disrupt or bypass “reason-responsiveness” will also be more mitigating than deficient self-control. On the other hand, the offender who is beset by overwhelming impulses to violence may well appear far more dangerous than the offender who wearily acquiesces to garden-variety impulses.

The mitigating effect of either mechanism described by Greenspan6’ - shortcutting or shortchanging - will likely be greater if the causal story begins, as it does in Caspi,& with the childhood maltreatment of the offender. Yet even a keen awareness of childhood victimization may

33

go only so far to mitigate our judgement of his adult violence. To the extent genetically vulnerable children become predisposed to violence by violent upbringings through such psychological mechanisms as modeling and identification, they will embrace the very attitudes and types of conduct that contributed to their own victimization. This may be poignant or ironic, but it would not, at least on account like Frankfurt’~,6~ make the victim-tuned-offender any less responsible.

CONCU~SION In the near future, human behavioral genetics may improve our ability to predict who will act violently in the kind of environments we expect to inhabit. But in the absence of sweeping institutional reform, and a deeper knowledge of mechanism than contemporary behavioral genetics can provide, that enhanced ability may be worse than useless. We have become highly skeptical, or cynical, about the old liberal nostrums of improved social services and dignified employment. But creating a more decent, just, and humane society would probably break the cycle of violence more appropriately, and perhaps more effec- tively, than any less comprehensive, more targeted intervention guided by behavioral genetics. On the other hand, behavioral genetics, in conjunction with sociology, developmental psychology, neurobiology, and cognitive science, may help us to understand why some people engage in destructive violence, and in understanding, to mitigate if not pardon their violent acts.

REPERANCES 1. N. Agar, in chapter 1 of ms. on political liberalism and

human genetic engineering, (London: Blackwell’s Publishing, forthcoming).

D. Wasserman, “Behavioral Genetics, Human” in TJ. MUTdy and M. J. Mehlman, eds., Encyclopedia of Ethical, Legal, and Policy Issues in Biotechnology (New York: John Wiley and Sons, 2000): 117-127; S.L. Sherman, J.C. DeFries, I. I. Gottesman, J.C. Loehlin, J.M. Meyer, M.Z. Pelias, J. Rice, I. Waldman, “Recent Developments in Human Behavioral Genetics: Past Accomplish- ments and Future Directions,” American Journal of Human Genetics 60 (1997): 1265-1275.

E. Balaban, “Reflections on Wye Woods: Crime, Biol- ogy, and Self-Interest,” Politics and the Life Sciences 15, no.1 (1996):86-88, at 87.

A. Caspi, J. McClay, T.E. Mofftt, J. Mill, J. Martin, I. W. Craig, A. Taylor, R. Poulton, “Role of Genotype in the Cycle of Violence in Maltreated Children,” Science 297 (August 2, 2002):

2.

3.

4.

851-854. 5. 6.

See Wasserman, supra note 2; at 123-126. D. Wasserman and R. Wachbroit, “Introduction: Mean-

ing, Methods, and Morals,” in D. Wasserman and R. Wachbroit, eds., Genetics and Criminal Behavior (New York: Cambridge University Press, 2001): 1-21, at 5-6.

K. Taylor, “On the Explanatory Limits of Behavioral Genetics,” in D. Wasserman and R. Wachbroit, eds., Genetics

7 .


32

and Criminal Behavior (New York: Cambridge University Press,

A. Gibbard, “Genetic Plans, Genetic Differences, and Violence: Some Chief Possibilities,” in D. Wasserman and R. Wachbroit, eds., Genetics and Criminal Behavior (New York: Cambridge University Press, 2001): 169-197, at 191.

See, e.g., Sherman, et. al., supra note 2.; R. Plomin, J.C. DeFries, G. McClearn, and M. Rutter, Behavioral Genetics, 3rd ed. (New York, Freeman, 1997).

See, e.g., G. Carey and I Gottesman, “Genetics and An- tisocial Behavior: Substance vs. Sound Bytes,” in Politics and the Life Sciences 15, no. 1 (1996) 88-90,

2001): 117-139. 8.

9.

10.

11. 12. Taylor, supra note 7. 13.

See, e.g., Sherman, et. al., supra note 2.

Features shared by virtually everyone in the population have little or no heritability; e.g., the trait of “having two a m ” will have very low heritability in a country where almost everyone does, and where most of those who don’t are missing limbs because of dismemberment or infection rather than genetic disease or mutation. It is important, then, to distinguish a trait’s heritability from its genetic basis, a distinction central to the general critique of behavioral genetics discussed earlier in the text. See E. Sober, “Separating Nature and Nurture,’’ in D. Wasserman and R. Wachbroit, eds., Genetics and Criminal Behavior (New York: Cambridge university Press, 2001): 47-78, at 55-56.

Sober, supra note 13, at 55, Sherman, et. al., supra note 2. C . Jencks, “Genes and Crime,” New YorkR- ofBooks

(February 12, 1987): 33-41, 34. See, eg., Sherman, supra note 2; Carey and Gottesman,

supra note 10. E. Balaban, “Human Correlative Behavioral Genetics:

An Alternative Viewpoint” in J. Benjamin, R. Ebstein, R. Belmaker, eds, Molecular Genetics and the Human Personality (Washing- ton, D. C.: American Psychiatric Publishing, 2002): 293-314, at

See, e.g., Sherman, et.al. supra note 2; Plomin, et, al.,

See, eg., Sherman, et. al., supra note 2. H.G. Brunner, M. Nelen, X. 0. Breakfield, H.H. Ropers,

and B.A. van Oost, “Abnormal Behavior Associated with a Point Mutation in the Structural Gene for Monoamine Oxidase,” Sci- ence 262 (193): 578-580.

E. Balaban, J. Alper, Y.L. Kasamon, “Mean Genes and the Biology of Human Aggression: A Critical Review of Recent Animal and Human Research,” Journalof Neurogenetics 11 (19%) 1-43, at 18.

I want to thank Dr. John Roberts for reminding me of this anomaly, in conversation at the 2003 Pitts Lectureship.

Caspi, et. al., s u p note 4. Caspi, et. al., supra note 4, at 852-853. Brunner et. al., supra note 20. A. Caspi, J. McClay, T.E. Moffitt, J. Mill, J. Martin, I. W.

Craig, A. Taylor, R. Poulton, “Description of Methods and Mea- surements Used in the Dunedin Multidisciplinary Health and Development Study,” supplementary material for “Role of Geno- type in the Cycle of Violence in Maltreated Children,” Science 297 (August 2, 2002): 851-854.

14. 15.

16.

17.

305-307. 18.

19. 20.

supra note 9.

21.

22.

23. 24. 25. 26.

27. 28. 29.

Caspi, et. al., supra note 4, at 853. Caspi, et. al., supra note 4 , at 852. Caspi, et. al., supra note 4, at 853. A year later, Caspi and

his colleagues, studying the same Dunedin cohort, reported an interestingly similar pattern of findings for the impact of life stress and variations of the serotonin transporter gene on depression. Life stress, like childhood maltreatment, had a main effect on

l%e Journal of Law, Medicine G Ethics

depression, while genetic variation on the transporter gene, as on the MAOA promoter gene, did not. There was again a significant gene x environment interaction, however, with life stress, like childhood maltreatment, associated with the phenotype - depression in the 2003 study - primarily in individuals with specific genetic variants - in the 2003 study, individuals with one or two copies of the short allele the promoter gene. Again, this suggests that individuals with that allele lacked the genetic “resilience” against environmental stress or trauma conferred by the other allele(s). A. Caspi, K. Sugden, T. E. Mofftt, A. Taylor, I. W. Craig, H. Harrington, J. McClay, J. Mill, J. Martin, A. Braithwaite, R. Poulton, “Influence of Life Stress on Depression: Moderation by a Polymorphism in the 5-HTT Gene,” Science 301 (2003): 386- 389.

30. 31. 32.

Caspi, et. al. supra note 4 ; Caspi, et al. supra note 26. Caspi, et. al. supra note 4. See, e.g., Improving the Well-Being of Abused and Ne-

glected children: Hearing Before the Senate Committee on Labor and Human resources, 104th Congress (1996): 9 , cited in R. Gordon, “Drifting through Byzantium: The Promise and Failure of the Adoption and Safe Families Act of 1997, ‘’ Minnesota Law Rm’m 83 (February 1999); 637-701, at 638.

Caspi, et. al. supra note 26, at 2. Maltreatment was con- stituted by such factors as negative or maternal affect, maternal inattention or unresponsiveness, and frequent changes in caregivers, which would far well below the legal threshold for neglect in most U.S. jurisdictions.

See, eg . , N. Abrams, “Problems in Defining Child Abuse and Neglect,” in W. Aiken and H. LaFollette, eds. Whose Child? Children’s Rghts, Paternal Authority, and State Power (Totowa, NJ: Rowman and Littlefield, 1980): 289-303, at 294-295; M. Guggenheim, “Somebody’s Child: Sustaining the Family’s Place in Child Welfare Policy [Book Review],” Harvard LawRevieu, 113 (2000): 1716-1750; D . Besharov, “Child Abuse Realities: Over- Reporting and Poverty,” mrginia Journal of Social Policy and Law 8 (2000): 165-203; M . Brining, “Choosing the Lesser Evil: Comments on Besharov’s ‘Child Abuse Realities’,” Virginia Jour- nal of Social Policy and Law 8 (2000): 205-218.

See, e.g., Besharov, supra note 34, at 190-192; D . Rob- erts, “The Challenge of Substance Abuse for Family Preservation Policy,” journal ofHealth Care Law and Policy 3 (1999): 72-87, 85-86; D. Lindsay, 7be Welfareof Children (1994): 28; L. H. Pelton, ForReasons of Poverty: A Critical Analysis of Child Welfare in the United States ( 1989).

E.g., Guggenheim, supra note 34 at 1728-1732. Caspi, et. al., supra note 26, at 4-5. The researchers coded maltreatment as “probable” if they

found only one indicator; as “severe” if they found two or more, Caspi, et. al., supra note 26, at 2-3.

Guggenheim, supra note 34, 1736-1738; Besharov, supra note 34, 199-202.

The phrase, and its context, are discussed in Wasserman and Wachbroit, supra note 6 at 5 .

Caspi, et. al. supra note 4, at 853. See, e.g., T. W. Adorno, E. Rrenkel-Brunswik, D.J.

Levinson. R.N. Sanford, 7be Authoritarian Pmonality: Studies in Prejudice (New York: Harper & Row, 1950).

33.

34.

35.

36. 37. 38.

39.

40.

41. 42.

43. 44.

45.

Caspi, et. al. supra note 4, at 852-853. See, e.g., the papers in this volume/for this symposium

by Franklin Zimring and Lance Stell. R. Wachbroit, “Genetic Determinism, Reductionism, and

Essentialism,” in T.J. Murray and M. J. Mehlman, eds., Encyclop-

dia of Ethical, Zkgal, and Policy Issues in Biotechnology (New York: John Wiley and Sons, 2000).

See, e.g., J.L.A. Garcia, “Strong Genetic Influence and the ‘New Optimism’ in D. Wasserman and R. Wachbroit, eds., Genetics and Criminal Behavior (New York: Cambridge Univer- sity Press, 2001): 273-302, at 273-274.

See, e.g, J.M. Fischer, “Recent Work in Moral Responsi- bility. Ethics 110, no. 1 (1999): 93-139.

See, e.g., S. J. Morse, “Uncontrollable Urges and Irratio- nal People,” Wqinia Law Review (forthcoming).

See, e.g., P. Greenspan, “Behavior Control and Freedom of Action,” Philosophical Review 87 (1978): 22-40.

H . Frankfurt, “Freedom of the Will and the Concept of a Person,” Journal of Philosophy 68 (1971): 5-20.

Accounts of moral responsibility based on these two features, which hardly exhaust the current versions of compatibilism, correspond roughly to what John Fisher, supra note 46 at 125-128, calls, respectively, “reason-responsiveness” and “identification” accounts.

46.

47.

48.

49.

50.

51.

52. 53. 54. 55.

Caspi, et. al. supra note 4, at 852-853. E.g., Morse, supra note 47. E.g., Greenspan, supra note 48. For broader skepticism about a criminal defense based

on a bad rearing environment, see R. Delgado, “’Rotten Social Background: Should the Criminal Law Recognize a Defense of Severe Environmental Deprivation?” Law and Inequality: A Jour- nal of7heoly and Practice 3 (1985): 9-90.

Caspi, et. al. supra note 4, at 853. See F. Schoeman, “Statistical Norms and Moral Attribu-

tions” in F. Schoeman, ed., Responsibility, Charactw, and the Emotions: New Essays in Moral psychobgy (Cambridge: Cambridge University Press, 1988) 287-315.

D. Wasserman, “Genetic Predispositions to Violent and Antisocial Behavior: Responsibility, Character, and Identity, Opti- mism” in D. Wasserman and R. Wachbroit, eds., Genetics and CriminalBehauior(New York: Cambridge University Press, 2001): 303-327, at 303-305.

59. G . Watson, “The Two Faces of Responsibility,” Philo- sophical Topics 24 (1996): 227-247.

60. Morse, supra note 47. 61.

62.

56. 57.

58.

T. Nagel, “Moral Luck,” Proceedings of the Aristotelian Society 50 (supplement) (1976): 137-152.

Caspi, et. al. supra note 4, at 853; for reviews of the research on serotonin, impulsivity, and violence, see e.g., D. Goldman, “The Search for Genetic Alleles Contributing to Self- Destructive and Aggressive Behaviors” in D. Stoff and R. Cairns, eds., Aggression and Violence: Genetic, Neurological, and Bioso- cia1 Penpectives (Mahwah, NJ: Erlbaum, 1996): 23-40.

P.S. Greenspan, “Genes, Electrotransmitters, and Free Will,” in D. Wasserman and R. Wachbroit, eds., Genetics and CriminalBebavior (New York: Cambridge University Press, 2001):

63.

243-258, 249. 64. Greenspan, supra note 62, at 249. 65. Frankfurt, supra at note 49. 66. J.M. Fischer and M. Rivazza, Responsibility and Control:

A 7heory of Moral Responsibility (Cambridge, Cambridge Univer- sity Press, 1988).

67. 68. 69.

Greenspan, supra note 62, at 249. Caspi, et. al., supra note 4 at 853. Frankfurt, supra at note 49.

33

Documents

Is There Value in Identifying Individual Genetic Predispositions to violence?